languages/LanguageConverter.php

   1 <?php
   2 /**
   3  * Contains the LanguageConverter class and ConverterRule class
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Language
  22  */
  23
  24 /**
  25  * Base class for language conversion.
  26  * @ingroup Language
  27  *
  28  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  29  * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>, PhiLiP <philip.npc@gmail.com>
  30  */
  31 class LanguageConverter {
  32
  33         /**
  34          * languages supporting variants
  35          * @since 1.20
  36          * @var array
  37          */
  38         static public $languagesWithVariants = array(
  39                 'gan',
  40                 'iu',
  41                 'kk',
  42                 'ku',
  43                 'shi',
  44                 'sr',
  45                 'tg',
  46                 'uz',
  47                 'zh',
  48         );
  49
  50         public $mMainLanguageCode;
  51         public $mVariants, $mVariantFallbacks, $mVariantNames;
  52         public $mTablesLoaded = false;
  53         public $mTables;
  54         // 'bidirectional' 'unidirectional' 'disable' for each variant
  55         public $mManualLevel;
  56
  57         /**
  58          * @var String: memcached key name
  59          */
  60         public $mCacheKey;
  61
  62         public $mLangObj;
  63         public $mFlags;
  64         public $mDescCodeSep = ':', $mDescVarSep = ';';
  65         public $mUcfirst = false;
  66         public $mConvRuleTitle = false;
  67         public $mURLVariant;
  68         public $mUserVariant;
  69         public $mHeaderVariant;
  70         public $mMaxDepth = 10;
  71         public $mVarSeparatorPattern;
  72
  73         const CACHE_VERSION_KEY = 'VERSION 7';
  74
  75         /**
  76          * Constructor
  77          *
  78          * @param $langobj Language: the Language Object
  79          * @param $maincode String: the main language code of this language
  80          * @param $variants Array: the supported variants of this language
  81          * @param $variantfallbacks Array: the fallback language of each variant
  82          * @param $flags Array: defining the custom strings that maps to the flags
  83          * @param $manualLevel Array: limit for supported variants
  84          */
  85         public function __construct( $langobj, $maincode, $variants = array(),
  86                                                                 $variantfallbacks = array(), $flags = array(),
  87                                                                 $manualLevel = array() ) {
  88                 global $wgDisabledVariants;
  89                 $this->mLangObj = $langobj;
  90                 $this->mMainLanguageCode = $maincode;
  91                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  92                 $this->mVariantFallbacks = $variantfallbacks;
  93                 $this->mVariantNames = Language::fetchLanguageNames();
  94                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  95                 $defaultflags = array(
  96                         // 'S' show converted text
  97                         // '+' add rules for alltext
  98                         // 'E' the gave flags is error
  99                         // these flags above are reserved for program
 100                         'A' => 'A',       // add rule for convert code (all text convert)
 101                         'T' => 'T',       // title convert
 102                         'R' => 'R',       // raw content
 103                         'D' => 'D',       // convert description (subclass implement)
 104                         '-' => '-',       // remove convert (not implement)
 105                         'H' => 'H',       // add rule for convert code
 106                                                   // (but no display in placed code)
 107                         'N' => 'N'        // current variant name
 108                 );
 109                 $this->mFlags = array_merge( $defaultflags, $flags );
 110                 foreach ( $this->mVariants as $v ) {
 111                         if ( array_key_exists( $v, $manualLevel ) ) {
 112                                 $this->mManualLevel[$v] = $manualLevel[$v];
 113                         } else {
 114                                 $this->mManualLevel[$v] = 'bidirectional';
 115                         }
 116                         $this->mFlags[$v] = $v;
 117                 }
 118         }
 119
 120         /**
 121          * Get all valid variants.
 122          * Call this instead of using $this->mVariants directly.
 123          *
 124          * @return Array: contains all valid variants
 125          */
 126         public function getVariants() {
 127                 return $this->mVariants;
 128         }
 129
 130         /**
 131          * In case some variant is not defined in the markup, we need
 132          * to have some fallback. For example, in zh, normally people
 133          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 134          * when zh-sg is preferred but not defined, we will pick zh-hans
 135          * in this case. Right now this is only used by zh.
 136          *
 137          * @param $variant String: the language code of the variant
 138          * @return String|array: The code of the fallback language or the
 139          *                               main code if there is no fallback
 140          */
 141         public function getVariantFallbacks( $variant ) {
 142                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 143                         return $this->mVariantFallbacks[$variant];
 144                 }
 145                 return $this->mMainLanguageCode;
 146         }
 147
 148         /**
 149          * Get the title produced by the conversion rule.
 150          * @return String: The converted title text
 151          */
 152         public function getConvRuleTitle() {
 153                 return $this->mConvRuleTitle;
 154         }
 155
 156         /**
 157          * Get preferred language variant.
 158          * @return String: the preferred language code
 159          */
 160         public function getPreferredVariant() {
 161                 global $wgDefaultLanguageVariant, $wgUser;
 162
 163                 $req = $this->getURLVariant();
 164
 165                 if ( $wgUser->isLoggedIn() && !$req ) {
 166                         $req = $this->getUserVariant();
 167                 } elseif ( !$req ) {
 168                         $req = $this->getHeaderVariant();
 169                 }
 170
 171                 if ( $wgDefaultLanguageVariant && !$req ) {
 172                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 173                 }
 174
 175                 // This function, unlike the other get*Variant functions, is
 176                 // not memoized (i.e. there return value is not cached) since
 177                 // new information might appear during processing after this
 178                 // is first called.
 179                 if ( $this->validateVariant( $req ) ) {
 180                         return $req;
 181                 }
 182                 return $this->mMainLanguageCode;
 183         }
 184
 185         /**
 186          * Get default variant.
 187          * This function would not be affected by user's settings
 188          * @return String: the default variant code
 189          */
 190         public function getDefaultVariant() {
 191                 global $wgDefaultLanguageVariant;
 192
 193                 $req = $this->getURLVariant();
 194
 195                 if ( !$req ) {
 196                         $req = $this->getHeaderVariant();
 197                 }
 198
 199                 if ( $wgDefaultLanguageVariant && !$req ) {
 200                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 201                 }
 202
 203                 if ( $req ) {
 204                         return $req;
 205                 }
 206                 return $this->mMainLanguageCode;
 207         }
 208
 209         /**
 210          * Validate the variant
 211          * @param $variant String: the variant to validate
 212          * @return Mixed: returns the variant if it is valid, null otherwise
 213          */
 214         public function validateVariant( $variant = null ) {
 215                 if ( $variant !== null && in_array( $variant, $this->mVariants ) ) {
 216                         return $variant;
 217                 }
 218                 return null;
 219         }
 220
 221         /**
 222          * Get the variant specified in the URL
 223          *
 224          * @return Mixed: variant if one found, false otherwise.
 225          */
 226         public function getURLVariant() {
 227                 global $wgRequest;
 228
 229                 if ( $this->mURLVariant ) {
 230                         return $this->mURLVariant;
 231                 }
 232
 233                 // see if the preference is set in the request
 234                 $ret = $wgRequest->getText( 'variant' );
 235
 236                 if ( !$ret ) {
 237                         $ret = $wgRequest->getVal( 'uselang' );
 238                 }
 239
 240                 return $this->mURLVariant = $this->validateVariant( $ret );
 241         }
 242
 243         /**
 244          * Determine if the user has a variant set.
 245          *
 246          * @return Mixed: variant if one found, false otherwise.
 247          */
 248         protected function getUserVariant() {
 249                 global $wgUser, $wgContLang;
 250
 251                 // memoizing this function wreaks havoc on parserTest.php
 252                 /*
 253                 if ( $this->mUserVariant ) {
 254                         return $this->mUserVariant;
 255                 }
 256                 */
 257
 258                 // Get language variant preference from logged in users
 259                 // Don't call this on stub objects because that causes infinite
 260                 // recursion during initialisation
 261                 if ( $wgUser->isLoggedIn() ) {
 262                         if ( $this->mMainLanguageCode == $wgContLang->getCode() ) {
 263                                 $ret = $wgUser->getOption( 'variant' );
 264                         } else {
 265                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 266                         }
 267                 } else {
 268                         // figure out user lang without constructing wgLang to avoid
 269                         // infinite recursion
 270                         $ret = $wgUser->getOption( 'language' );
 271                 }
 272
 273                 return $this->mUserVariant = $this->validateVariant( $ret );
 274         }
 275
 276         /**
 277          * Determine the language variant from the Accept-Language header.
 278          *
 279          * @return Mixed: variant if one found, false otherwise.
 280          */
 281         protected function getHeaderVariant() {
 282                 global $wgRequest;
 283
 284                 if ( $this->mHeaderVariant ) {
 285                         return $this->mHeaderVariant;
 286                 }
 287
 288                 // see if some supported language variant is set in the
 289                 // HTTP header.
 290                 $languages = array_keys( $wgRequest->getAcceptLang() );
 291                 if ( empty( $languages ) ) {
 292                         return null;
 293                 }
 294
 295                 $fallbackLanguages = array();
 296                 foreach ( $languages as $language ) {
 297                         $this->mHeaderVariant = $this->validateVariant( $language );
 298                         if ( $this->mHeaderVariant ) {
 299                                 break;
 300                         }
 301
 302                         // To see if there are fallbacks of current language.
 303                         // We record these fallback variants, and process
 304                         // them later.
 305                         $fallbacks = $this->getVariantFallbacks( $language );
 306                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 307                                 $fallbackLanguages[] = $fallbacks;
 308                         } elseif ( is_array( $fallbacks ) ) {
 309                                 $fallbackLanguages =
 310                                         array_merge( $fallbackLanguages, $fallbacks );
 311                         }
 312                 }
 313
 314                 if ( !$this->mHeaderVariant ) {
 315                         // process fallback languages now
 316                         $fallback_languages = array_unique( $fallbackLanguages );
 317                         foreach ( $fallback_languages as $language ) {
 318                                 $this->mHeaderVariant = $this->validateVariant( $language );
 319                                 if ( $this->mHeaderVariant ) {
 320                                         break;
 321                                 }
 322                         }
 323                 }
 324
 325                 return $this->mHeaderVariant;
 326         }
 327
 328         /**
 329          * Dictionary-based conversion.
 330          * This function would not parse the conversion rules.
 331          * If you want to parse rules, try to use convert() or
 332          * convertTo().
 333          *
 334          * @param $text String the text to be converted
 335          * @param $toVariant bool|string the target language code
 336          * @return String the converted text
 337          */
 338         public function autoConvert( $text, $toVariant = false ) {
 339                 wfProfileIn( __METHOD__ );
 340
 341                 $this->loadTables();
 342
 343                 if ( !$toVariant ) {
 344                         $toVariant = $this->getPreferredVariant();
 345                         if ( !$toVariant ) {
 346                                 wfProfileOut( __METHOD__ );
 347                                 return $text;
 348                         }
 349                 }
 350
 351                 if ( $this->guessVariant( $text, $toVariant ) ) {
 352                         wfProfileOut( __METHOD__ );
 353                         return $text;
 354                 }
 355
 356                 /* we convert everything except:
 357                    1. HTML markups (anything between < and >)
 358                    2. HTML entities
 359                    3. placeholders created by the parser
 360                 */
 361                 global $wgParser;
 362                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 363                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 364                 } else {
 365                         $marker = '';
 366                 }
 367
 368                 // this one is needed when the text is inside an HTML markup
 369                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 370
 371                 // disable convert to variants between <code> tags
 372                 $codefix = '<code>.+?<\/code>|';
 373                 // disable conversion of <script> tags
 374                 $scriptfix = '<script.*?>.*?<\/script>|';
 375                 // disable conversion of <pre> tags
 376                 $prefix = '<pre.*?>.*?<\/pre>|';
 377
 378                 $reg = '/' . $codefix . $scriptfix . $prefix .
 379                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 380                 $startPos = 0;
 381                 $sourceBlob = '';
 382                 $literalBlob = '';
 383
 384                 // Guard against delimiter nulls in the input
 385                 $text = str_replace( "\000", '', $text );
 386
 387                 $markupMatches = null;
 388                 $elementMatches = null;
 389                 while ( $startPos < strlen( $text ) ) {
 390                         if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 391                                 $elementPos = $markupMatches[0][1];
 392                                 $element = $markupMatches[0][0];
 393                         } else {
 394                                 $elementPos = strlen( $text );
 395                                 $element = '';
 396                         }
 397
 398                         // Queue the part before the markup for translation in a batch
 399                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 400
 401                         // Advance to the next position
 402                         $startPos = $elementPos + strlen( $element );
 403
 404                         // Translate any alt or title attributes inside the matched element
 405                         if ( $element !== '' && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element,
 406                                 $elementMatches ) )
 407                         {
 408                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 409                                 $changed = false;
 410                                 foreach ( array( 'title', 'alt' ) as $attrName ) {
 411                                         if ( !isset( $attrs[$attrName] ) ) {
 412                                                 continue;
 413                                         }
 414                                         $attr = $attrs[$attrName];
 415                                         // Don't convert URLs
 416                                         if ( !strpos( $attr, '://' ) ) {
 417                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 418                                         }
 419
 420                                         // Remove HTML tags to avoid disrupting the layout
 421                                         $attr = preg_replace( '/<[^>]+>/', '', $attr );
 422                                         if ( $attr !== $attrs[$attrName] ) {
 423                                                 $attrs[$attrName] = $attr;
 424                                                 $changed = true;
 425                                         }
 426                                 }
 427                                 if ( $changed ) {
 428                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 429                                                 $elementMatches[3];
 430                                 }
 431                         }
 432                         $literalBlob .= $element . "\000";
 433                 }
 434
 435                 // Do the main translation batch
 436                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 437
 438                 // Put the output back together
 439                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 440                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 441                 $output = '';
 442                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 443                         $output .= $translatedIter->current();
 444                         $output .= $literalIter->current();
 445                         $translatedIter->next();
 446                         $literalIter->next();
 447                 }
 448
 449                 wfProfileOut( __METHOD__ );
 450                 return $output;
 451         }
 452
 453         /**
 454          * Translate a string to a variant.
 455          * Doesn't parse rules or do any of that other stuff, for that use
 456          * convert() or convertTo().
 457          *
 458          * @param $text String: text to convert
 459          * @param $variant String: variant language code
 460          * @return String: translated text
 461          */
 462         public function translate( $text, $variant ) {
 463                 wfProfileIn( __METHOD__ );
 464                 // If $text is empty or only includes spaces, do nothing
 465                 // Otherwise translate it
 466                 if ( trim( $text ) ) {
 467                         $this->loadTables();
 468                         $text = $this->mTables[$variant]->replace( $text );
 469                 }
 470                 wfProfileOut( __METHOD__ );
 471                 return $text;
 472         }
 473
 474         /**
 475          * Call translate() to convert text to all valid variants.
 476          *
 477          * @param $text String: the text to be converted
 478          * @return Array: variant => converted text
 479          */
 480         public function autoConvertToAllVariants( $text ) {
 481                 wfProfileIn( __METHOD__ );
 482                 $this->loadTables();
 483
 484                 $ret = array();
 485                 foreach ( $this->mVariants as $variant ) {
 486                         $ret[$variant] = $this->translate( $text, $variant );
 487                 }
 488
 489                 wfProfileOut( __METHOD__ );
 490                 return $ret;
 491         }
 492
 493         /**
 494          * Convert link text to all valid variants.
 495          * In the first, this function only convert text outside the
 496          * "-{" "}-" markups. Since the "{" and "}" are not allowed in
 497          * titles, the text will get all converted always.
 498          * So I removed this feature and deprecated the function.
 499          *
 500          * @param $text String: the text to be converted
 501          * @return Array: variant => converted text
 502          * @deprecated since 1.17 Use autoConvertToAllVariants() instead
 503          */
 504         public function convertLinkToAllVariants( $text ) {
 505                 return $this->autoConvertToAllVariants( $text );
 506         }
 507
 508         /**
 509          * Apply manual conversion rules.
 510          *
 511          * @param $convRule ConverterRule Object of ConverterRule
 512          */
 513         protected function applyManualConv( $convRule ) {
 514                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 515                 // title conversion.
 516                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 517                 // rule(s) not for title, this breaks the title conversion.
 518                 $newConvRuleTitle = $convRule->getTitle();
 519                 if ( $newConvRuleTitle ) {
 520                         // So I add an empty check for getTitle()
 521                         $this->mConvRuleTitle = $newConvRuleTitle;
 522                 }
 523
 524                 // merge/remove manual conversion rules to/from global table
 525                 $convTable = $convRule->getConvTable();
 526                 $action = $convRule->getRulesAction();
 527                 foreach ( $convTable as $variant => $pair ) {
 528                         if ( !$this->validateVariant( $variant ) ) {
 529                                 continue;
 530                         }
 531
 532                         if ( $action == 'add' ) {
 533                                 foreach ( $pair as $from => $to ) {
 534                                         // to ensure that $from and $to not be left blank
 535                                         // so $this->translate() could always return a string
 536                                         if ( $from || $to ) {
 537                                                 // more efficient than array_merge(), about 2.5 times.
 538                                                 $this->mTables[$variant]->setPair( $from, $to );
 539                                         }
 540                                 }
 541                         } elseif ( $action == 'remove' ) {
 542                                 $this->mTables[$variant]->removeArray( $pair );
 543                         }
 544                 }
 545         }
 546
 547         /**
 548          * Auto convert a Title object to a readable string in the
 549          * preferred variant.
 550          *
 551          * @param $title Title a object of Title
 552          * @return String: converted title text
 553          */
 554         public function convertTitle( $title ) {
 555                 $variant = $this->getPreferredVariant();
 556                 $index = $title->getNamespace();
 557                 if ( $index !== NS_MAIN ) {
 558                         $text = $this->convertNamespace( $index, $variant ) . ':';
 559                 } else {
 560                         $text = '';
 561                 }
 562                 $text .= $this->translate( $title->getText(), $variant );
 563                 return $text;
 564         }
 565
 566         /**
 567          * Get the namespace display name in the preferred variant.
 568          *
 569          * @param $index int namespace id
 570          * @param $variant string|null variant code or null for preferred variant
 571          * @return String: namespace name for display
 572          */
 573         public function convertNamespace( $index, $variant = null ) {
 574                 if ( $variant === null ) {
 575                         $variant = $this->getPreferredVariant();
 576                 }
 577                 if ( $index === NS_MAIN ) {
 578                         return '';
 579                 } else {
 580                         // First check if a message gives a converted name in the target variant.
 581                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 582                         if ( $nsConvMsg->exists() ) {
 583                                 return $nsConvMsg->plain();
 584                         }
 585                         // Then check if a message gives a converted name in content language
 586                         // which needs extra translation to the target variant.
 587                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 588                         if ( $nsConvMsg->exists() ) {
 589                                 return $this->translate( $nsConvMsg->plain(), $variant );
 590                         }
 591                         // No message exists, retrieve it from the target variant's namespace names.
 592                         $langObj = $this->mLangObj->factory( $variant );
 593                         return $langObj->getFormattedNsText( $index );
 594                 }
 595         }
 596
 597         /**
 598          * Convert text to different variants of a language. The automatic
 599          * conversion is done in autoConvert(). Here we parse the text
 600          * marked with -{}-, which specifies special conversions of the
 601          * text that can not be accomplished in autoConvert().
 602          *
 603          * Syntax of the markup:
 604          * -{code1:text1;code2:text2;...}-  or
 605          * -{flags|code1:text1;code2:text2;...}-  or
 606          * -{text}- in which case no conversion should take place for text
 607          *
 608          * @param $text String: text to be converted
 609          * @return String: converted text
 610          */
 611         public function convert( $text ) {
 612                 $variant = $this->getPreferredVariant();
 613                 return $this->convertTo( $text, $variant );
 614         }
 615
 616         /**
 617          * Same as convert() except a extra parameter to custom variant.
 618          *
 619          * @param $text String: text to be converted
 620          * @param $variant String: the target variant code
 621          * @return String: converted text
 622          */
 623         public function convertTo( $text, $variant ) {
 624                 global $wgDisableLangConversion;
 625                 if ( $wgDisableLangConversion ) {
 626                         return $text;
 627                 }
 628                 // Reset converter state for a new converter run.
 629                 $this->mConvRuleTitle = false;
 630                 return $this->recursiveConvertTopLevel( $text, $variant );
 631         }
 632
 633         /**
 634          * Recursively convert text on the outside. Allow to use nested
 635          * markups to custom rules.
 636          *
 637          * @param $text String: text to be converted
 638          * @param $variant String: the target variant code
 639          * @param $depth Integer: depth of recursion
 640          * @return String: converted text
 641          */
 642         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 643                 $startPos = 0;
 644                 $out = '';
 645                 $length = strlen( $text );
 646                 $shouldConvert = !$this->guessVariant( $text, $variant );
 647
 648                 while ( $startPos < $length ) {
 649                         $pos = strpos( $text, '-{', $startPos );
 650
 651                         if ( $pos === false ) {
 652                                 // No more markup, append final segment
 653                                 $fragment = substr( $text, $startPos );
 654                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 655                                 return $out;
 656                         }
 657
 658                         // Markup found
 659                         // Append initial segment
 660                         $fragment = substr( $text, $startPos, $pos - $startPos );
 661                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 662
 663                         // Advance position
 664                         $startPos = $pos;
 665
 666                         // Do recursive conversion
 667                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 668                 }
 669
 670                 return $out;
 671         }
 672
 673         /**
 674          * Recursively convert text on the inside.
 675          *
 676          * @param $text String: text to be converted
 677          * @param $variant String: the target variant code
 678          * @param $startPos int
 679          * @param $depth Integer: depth of recursion
 680          *
 681          * @throws MWException
 682          * @return String: converted text
 683          */
 684         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 685                 // Quick sanity check (no function calls)
 686                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 687                         throw new MWException( __METHOD__ . ': invalid input string' );
 688                 }
 689
 690                 $startPos += 2;
 691                 $inner = '';
 692                 $warningDone = false;
 693                 $length = strlen( $text );
 694
 695                 while ( $startPos < $length ) {
 696                         $m = false;
 697                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 698                         if ( !$m ) {
 699                                 // Unclosed rule
 700                                 break;
 701                         }
 702
 703                         $token = $m[0][0];
 704                         $pos = $m[0][1];
 705
 706                         // Markup found
 707                         // Append initial segment
 708                         $inner .= substr( $text, $startPos, $pos - $startPos );
 709
 710                         // Advance position
 711                         $startPos = $pos;
 712
 713                         switch ( $token ) {
 714                                 case '-{':
 715                                         // Check max depth
 716                                         if ( $depth >= $this->mMaxDepth ) {
 717                                                 $inner .= '-{';
 718                                                 if ( !$warningDone ) {
 719                                                         $inner .= '<span class="error">' .
 720                                                                 wfMessage( 'language-converter-depth-warning' )
 721                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 722                                                                 '</span>';
 723                                                         $warningDone = true;
 724                                                 }
 725                                                 $startPos += 2;
 726                                                 continue;
 727                                         }
 728                                         // Recursively parse another rule
 729                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 730                                         break;
 731                                 case '}-':
 732                                         // Apply the rule
 733                                         $startPos += 2;
 734                                         $rule = new ConverterRule( $inner, $this );
 735                                         $rule->parse( $variant );
 736                                         $this->applyManualConv( $rule );
 737                                         return $rule->getDisplay();
 738                                 default:
 739                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 740                         }
 741                 }
 742
 743                 // Unclosed rule
 744                 if ( $startPos < $length ) {
 745                         $inner .= substr( $text, $startPos );
 746                 }
 747                 $startPos = $length;
 748                 return '-{' . $this->autoConvert( $inner, $variant );
 749         }
 750
 751         /**
 752          * If a language supports multiple variants, it is possible that
 753          * non-existing link in one variant actually exists in another variant.
 754          * This function tries to find it. See e.g. LanguageZh.php
 755          *
 756          * @param $link String: the name of the link
 757          * @param $nt Mixed: the title object of the link
 758          * @param $ignoreOtherCond Boolean: to disable other conditions when
 759          *              we need to transclude a template or update a category's link
 760          * @return Null, the input parameters may be modified upon return
 761          */
 762         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 763                 # If the article has already existed, there is no need to
 764                 # check it again, otherwise it may cause a fault.
 765                 if ( is_object( $nt ) && $nt->exists() ) {
 766                         return;
 767                 }
 768
 769                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 770                         $wgUser;
 771                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 772                 $action = $wgRequest->getText( 'action' );
 773                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 774                 $disableLinkConversion = $wgDisableLangConversion
 775                         || $wgDisableTitleConversion;
 776                 $linkBatch = new LinkBatch();
 777
 778                 $ns = NS_MAIN;
 779
 780                 if ( $disableLinkConversion ||
 781                          ( !$ignoreOtherCond &&
 782                            ( $isredir == 'no'
 783                                  || $action == 'edit'
 784                                  || $action == 'submit'
 785                                  || $linkconvert == 'no'
 786                                  || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 787                         return;
 788                 }
 789
 790                 if ( is_object( $nt ) ) {
 791                         $ns = $nt->getNamespace();
 792                 }
 793
 794                 $variants = $this->autoConvertToAllVariants( $link );
 795                 if ( !$variants ) { // give up
 796                         return;
 797                 }
 798
 799                 $titles = array();
 800
 801                 foreach ( $variants as $v ) {
 802                         if ( $v != $link ) {
 803                                 $varnt = Title::newFromText( $v, $ns );
 804                                 if ( !is_null( $varnt ) ) {
 805                                         $linkBatch->addObj( $varnt );
 806                                         $titles[] = $varnt;
 807                                 }
 808                         }
 809                 }
 810
 811                 // fetch all variants in single query
 812                 $linkBatch->execute();
 813
 814                 foreach ( $titles as $varnt ) {
 815                         if ( $varnt->getArticleID() > 0 ) {
 816                                 $nt = $varnt;
 817                                 $link = $varnt->getText();
 818                                 break;
 819                         }
 820                 }
 821         }
 822
 823         /**
 824          * Returns language specific hash options.
 825          *
 826          * @return string
 827          */
 828         public function getExtraHashOptions() {
 829                 $variant = $this->getPreferredVariant();
 830                 return '!' . $variant;
 831         }
 832
 833         /**
 834          * Guess if a text is written in a variant. This should be implemented in subclasses.
 835          *
 836          * @param string $text the text to be checked
 837          * @param string $variant language code of the variant to be checked for
 838          * @return bool true if $text appears to be written in $variant, false if not
 839          *
 840          * @author Nikola Smolenski <smolensk@eunet.rs>
 841          * @since 1.19
 842          */
 843         public function guessVariant( $text, $variant ) {
 844                 return false;
 845         }
 846
 847         /**
 848          * Load default conversion tables.
 849          * This method must be implemented in derived class.
 850          *
 851          * @private
 852          * @throws MWException
 853          */
 854         function loadDefaultTables() {
 855                 $name = get_class( $this );
 856                 throw new MWException( "Must implement loadDefaultTables() method in class $name" );
 857         }
 858
 859         /**
 860          * Load conversion tables either from the cache or the disk.
 861          * @private
 862          * @param $fromCache Boolean: load from memcached? Defaults to true.
 863          */
 864         function loadTables( $fromCache = true ) {
 865                 global $wgLangConvMemc;
 866
 867                 if ( $this->mTablesLoaded ) {
 868                         return;
 869                 }
 870
 871                 wfProfileIn( __METHOD__ );
 872                 $this->mTablesLoaded = true;
 873                 $this->mTables = false;
 874                 if ( $fromCache ) {
 875                         wfProfileIn( __METHOD__ . '-cache' );
 876                         $this->mTables = $wgLangConvMemc->get( $this->mCacheKey );
 877                         wfProfileOut( __METHOD__ . '-cache' );
 878                 }
 879                 if ( !$this->mTables
 880                          || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 881                         wfProfileIn( __METHOD__ . '-recache' );
 882                         // not in cache, or we need a fresh reload.
 883                         // We will first load the default tables
 884                         // then update them using things in MediaWiki:Conversiontable/*
 885                         $this->loadDefaultTables();
 886                         foreach ( $this->mVariants as $var ) {
 887                                 $cached = $this->parseCachedTable( $var );
 888                                 $this->mTables[$var]->mergeArray( $cached );
 889                         }
 890
 891                         $this->postLoadTables();
 892                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 893
 894                         $wgLangConvMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 895                         wfProfileOut( __METHOD__ . '-recache' );
 896                 }
 897                 wfProfileOut( __METHOD__ );
 898         }
 899
 900         /**
 901          * Hook for post processing after conversion tables are loaded.
 902          */
 903         function postLoadTables() { }
 904
 905         /**
 906          * Reload the conversion tables.
 907          *
 908          * @private
 909          */
 910         function reloadTables() {
 911                 if ( $this->mTables ) {
 912                         unset( $this->mTables );
 913                 }
 914                 $this->mTablesLoaded = false;
 915                 $this->loadTables( false );
 916         }
 917
 918         /**
 919          * Parse the conversion table stored in the cache.
 920          *
 921          * The tables should be in blocks of the following form:
 922          *              -{
 923          *                      word => word ;
 924          *                      word => word ;
 925          *                      ...
 926          *              }-
 927          *
 928          * To make the tables more manageable, subpages are allowed
 929          * and will be parsed recursively if $recursive == true.
 930          *
 931          * @param $code String: language code
 932          * @param $subpage String: subpage name
 933          * @param $recursive Boolean: parse subpages recursively? Defaults to true.
 934          *
 935          * @return array
 936          */
 937         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 938                 static $parsed = array();
 939
 940                 $key = 'Conversiontable/' . $code;
 941                 if ( $subpage ) {
 942                         $key .= '/' . $subpage;
 943                 }
 944                 if ( array_key_exists( $key, $parsed ) ) {
 945                         return array();
 946                 }
 947
 948                 $parsed[$key] = true;
 949
 950                 if ( $subpage === '' ) {
 951                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
 952                 } else {
 953                         $txt = false;
 954                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 955                         if ( $title && $title->exists() ) {
 956                                 $revision = Revision::newFromTitle( $title );
 957                                 if ( $revision ) {
 958                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
 959                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
 960                                         }
 961
 962                                         // @todo in the future, use a specialized content model, perhaps based on json!
 963                                 }
 964                         }
 965                 }
 966
 967                 # Nothing to parse if there's no text
 968                 if ( $txt === false || $txt === null || $txt === '' ) {
 969                         return array();
 970                 }
 971
 972                 // get all subpage links of the form
 973                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 974                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 975                         ':Conversiontable';
 976                 $subs = StringUtils::explode( '[[', $txt );
 977                 $sublinks = array();
 978                 foreach ( $subs as $sub ) {
 979                         $link = explode( ']]', $sub, 2 );
 980                         if ( count( $link ) != 2 ) {
 981                                 continue;
 982                         }
 983                         $b = explode( '|', $link[0], 2 );
 984                         $b = explode( '/', trim( $b[0] ), 3 );
 985                         if ( count( $b ) == 3 ) {
 986                                 $sublink = $b[2];
 987                         } else {
 988                                 $sublink = '';
 989                         }
 990
 991                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 992                                 $sublinks[] = $sublink;
 993                         }
 994                 }
 995
 996                 // parse the mappings in this page
 997                 $blocks = StringUtils::explode( '-{', $txt );
 998                 $ret = array();
 999                 $first = true;
1000                 foreach ( $blocks as $block ) {
1001                         if ( $first ) {
1002                                 // Skip the part before the first -{
1003                                 $first = false;
1004                                 continue;
1005                         }
1006                         $mappings = explode( '}-', $block, 2 );
1007                         $stripped = str_replace( array( "'", '"', '*', '#' ), '',
1008                                                                          $mappings[0] );
1009                         $table = StringUtils::explode( ';', $stripped );
1010                         foreach ( $table as $t ) {
1011                                 $m = explode( '=>', $t, 3 );
1012                                 if ( count( $m ) != 2 ) {
1013                                         continue;
1014                                 }
1015                                 // trim any trailling comments starting with '//'
1016                                 $tt = explode( '//', $m[1], 2 );
1017                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1018                         }
1019                 }
1020
1021                 // recursively parse the subpages
1022                 if ( $recursive ) {
1023                         foreach ( $sublinks as $link ) {
1024                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1025                                 $ret = array_merge( $ret, $s );
1026                         }
1027                 }
1028
1029                 if ( $this->mUcfirst ) {
1030                         foreach ( $ret as $k => $v ) {
1031                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1032                         }
1033                 }
1034                 return $ret;
1035         }
1036
1037         /**
1038          * Enclose a string with the "no conversion" tag. This is used by
1039          * various functions in the Parser.
1040          *
1041          * @param $text String: text to be tagged for no conversion
1042          * @param $noParse Boolean: unused
1043          * @return String: the tagged text
1044          */
1045         public function markNoConversion( $text, $noParse = false ) {
1046                 # don't mark if already marked
1047                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1048                         return $text;
1049                 }
1050
1051                 $ret = "-{R|$text}-";
1052                 return $ret;
1053         }
1054
1055         /**
1056          * Convert the sorting key for category links. This should make different
1057          * keys that are variants of each other map to the same key.
1058          *
1059          * @param $key string
1060          *
1061          * @return string
1062          */
1063         function convertCategoryKey( $key ) {
1064                 return $key;
1065         }
1066
1067         /**
1068          * Hook to refresh the cache of conversion tables when
1069          * MediaWiki:Conversiontable* is updated.
1070          * @private
1071          *
1072          * @param $page WikiPage object
1073          * @param $user Object: User object for the current user
1074          * @param $content Content: new page content
1075          * @param $summary String: edit summary of the edit
1076          * @param $isMinor Boolean: was the edit marked as minor?
1077          * @param $isWatch Boolean: did the user watch this page or not?
1078          * @param $section
1079          * @param $flags int Bitfield
1080          * @param $revision Object: new Revision object or null
1081          * @return Boolean: true
1082          */
1083         function OnPageContentSaveComplete( $page, $user, $content, $summary, $isMinor,
1084                         $isWatch, $section, $flags, $revision ) {
1085                 $titleobj = $page->getTitle();
1086                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1087                         $title = $titleobj->getDBkey();
1088                         $t = explode( '/', $title, 3 );
1089                         $c = count( $t );
1090                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1091                                 if ( $this->validateVariant( $t[1] ) ) {
1092                                         $this->reloadTables();
1093                                 }
1094                         }
1095                 }
1096                 return true;
1097         }
1098
1099         /**
1100          * Armour rendered math against conversion.
1101          * Escape special chars in parsed math text. (in most cases are img elements)
1102          *
1103          * @param $text String: text to armour against conversion
1104          * @return String: armoured text where { and } have been converted to
1105          *                 &#123; and &#125;
1106          * @deprecated since 1.22 is no longer used
1107          */
1108         public function armourMath( $text ) {
1109                 // convert '-{' and '}-' to '-&#123;' and '&#125;-' to prevent
1110                 // any unwanted markup appearing in the math image tag.
1111                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
1112                 return $text;
1113         }
1114
1115         /**
1116          * Get the cached separator pattern for ConverterRule::parseRules()
1117          */
1118         function getVarSeparatorPattern() {
1119                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1120                         // varsep_pattern for preg_split:
1121                         // text should be splited by ";" only if a valid variant
1122                         // name exist after the markup, for example:
1123                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1124                         //      <span style="font-size:120%;">yyy</span>;}-
1125                         // we should split it as:
1126                         //  array(
1127                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1128                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1129                         //        [2] => ''
1130                         //       )
1131                         $pat = '/;\s*(?=';
1132                         foreach ( $this->mVariants as $variant ) {
1133                                 // zh-hans:xxx;zh-hant:yyy
1134                                 $pat .= $variant . '\s*:|';
1135                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1136                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1137                         }
1138                         $pat .= '\s*$)/';
1139                         $this->mVarSeparatorPattern = $pat;
1140                 }
1141                 return $this->mVarSeparatorPattern;
1142         }
1143 }
1144
1145 /**
1146  * Parser for rules of language conversion , parse rules in -{ }- tag.
1147  * @ingroup Language
1148  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1149  */
1150 class ConverterRule {
1151         public $mText; // original text in -{text}-
1152         public $mConverter; // LanguageConverter object
1153         public $mRuleDisplay = '';
1154         public $mRuleTitle = false;
1155         public $mRules = '';// string : the text of the rules
1156         public $mRulesAction = 'none';
1157         public $mFlags = array();
1158         public $mVariantFlags = array();
1159         public $mConvTable = array();
1160         public $mBidtable = array();// array of the translation in each variant
1161         public $mUnidtable = array();// array of the translation in each variant
1162
1163         /**
1164          * Constructor
1165          *
1166          * @param $text String: the text between -{ and }-
1167          * @param $converter LanguageConverter object
1168          */
1169         public function __construct( $text, $converter ) {
1170                 $this->mText = $text;
1171                 $this->mConverter = $converter;
1172         }
1173
1174         /**
1175          * Check if variants array in convert array.
1176          *
1177          * @param $variants Array or string: variant language code
1178          * @return String: translated text
1179          */
1180         public function getTextInBidtable( $variants ) {
1181                 $variants = (array)$variants;
1182                 if ( !$variants ) {
1183                         return false;
1184                 }
1185                 foreach ( $variants as $variant ) {
1186                         if ( isset( $this->mBidtable[$variant] ) ) {
1187                                 return $this->mBidtable[$variant];
1188                         }
1189                 }
1190                 return false;
1191         }
1192
1193         /**
1194          * Parse flags with syntax -{FLAG| ... }-
1195          * @private
1196          */
1197         function parseFlags() {
1198                 $text = $this->mText;
1199                 $flags = array();
1200                 $variantFlags = array();
1201
1202                 $sepPos = strpos( $text, '|' );
1203                 if ( $sepPos !== false ) {
1204                         $validFlags = $this->mConverter->mFlags;
1205                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1206                         foreach ( $f as $ff ) {
1207                                 $ff = trim( $ff );
1208                                 if ( isset( $validFlags[$ff] ) ) {
1209                                         $flags[$validFlags[$ff]] = true;
1210                                 }
1211                         }
1212                         $text = strval( substr( $text, $sepPos + 1 ) );
1213                 }
1214
1215                 if ( !$flags ) {
1216                         $flags['S'] = true;
1217                 } elseif ( isset( $flags['R'] ) ) {
1218                         $flags = array( 'R' => true );// remove other flags
1219                 } elseif ( isset( $flags['N'] ) ) {
1220                         $flags = array( 'N' => true );// remove other flags
1221                 } elseif ( isset( $flags['-'] ) ) {
1222                         $flags = array( '-' => true );// remove other flags
1223                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1224                         $flags['H'] = true;
1225                 } elseif ( isset( $flags['H'] ) ) {
1226                         // replace A flag, and remove other flags except T
1227                         $temp = array( '+' => true, 'H' => true );
1228                         if ( isset( $flags['T'] ) ) {
1229                                 $temp['T'] = true;
1230                         }
1231                         if ( isset( $flags['D'] ) ) {
1232                                 $temp['D'] = true;
1233                         }
1234                         $flags = $temp;
1235                 } else {
1236                         if ( isset( $flags['A'] ) ) {
1237                                 $flags['+'] = true;
1238                                 $flags['S'] = true;
1239                         }
1240                         if ( isset( $flags['D'] ) ) {
1241                                 unset( $flags['S'] );
1242                         }
1243                         // try to find flags like "zh-hans", "zh-hant"
1244                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1245                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1246                         if ( $variantFlags ) {
1247                                 $variantFlags = array_flip( $variantFlags );
1248                                 $flags = array();
1249                         }
1250                 }
1251                 $this->mVariantFlags = $variantFlags;
1252                 $this->mRules = $text;
1253                 $this->mFlags = $flags;
1254         }
1255
1256         /**
1257          * Generate conversion table.
1258          * @private
1259          */
1260         function parseRules() {
1261                 $rules = $this->mRules;
1262                 $bidtable = array();
1263                 $unidtable = array();
1264                 $variants = $this->mConverter->mVariants;
1265                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1266
1267                 // Split according to $varsep_pattern, but ignore semicolons from HTML entities
1268                 $rules = preg_replace( '/(&[#a-zA-Z0-9]+);/', "$1\x01", $rules );
1269                 $choice = preg_split( $varsep_pattern, $rules );
1270                 $choice = str_replace( "\x01", ';', $choice );
1271
1272                 foreach ( $choice as $c ) {
1273                         $v = explode( ':', $c, 2 );
1274                         if ( count( $v ) != 2 ) {
1275                                 // syntax error, skip
1276                                 continue;
1277                         }
1278                         $to = trim( $v[1] );
1279                         $v = trim( $v[0] );
1280                         $u = explode( '=>', $v, 2 );
1281                         // if $to is empty, strtr() could return a wrong result
1282                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1283                                 $bidtable[$v] = $to;
1284                         } elseif ( count( $u ) == 2 ) {
1285                                 $from = trim( $u[0] );
1286                                 $v = trim( $u[1] );
1287                                 if ( array_key_exists( $v, $unidtable )
1288                                          && !is_array( $unidtable[$v] )
1289                                          && $to
1290                                          && in_array( $v, $variants ) ) {
1291                                         $unidtable[$v] = array( $from => $to );
1292                                 } elseif ( $to && in_array( $v, $variants ) ) {
1293                                         $unidtable[$v][$from] = $to;
1294                                 }
1295                         }
1296                         // syntax error, pass
1297                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1298                                 $bidtable = array();
1299                                 $unidtable = array();
1300                                 break;
1301                         }
1302                 }
1303                 $this->mBidtable = $bidtable;
1304                 $this->mUnidtable = $unidtable;
1305         }
1306
1307         /**
1308          * @private
1309          *
1310          * @return string
1311          */
1312         function getRulesDesc() {
1313                 $codesep = $this->mConverter->mDescCodeSep;
1314                 $varsep = $this->mConverter->mDescVarSep;
1315                 $text = '';
1316                 foreach ( $this->mBidtable as $k => $v ) {
1317                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1318                 }
1319                 foreach ( $this->mUnidtable as $k => $a ) {
1320                         foreach ( $a as $from => $to ) {
1321                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1322                                         "$codesep$to$varsep";
1323                         }
1324                 }
1325                 return $text;
1326         }
1327
1328         /**
1329          * Parse rules conversion.
1330          * @private
1331          *
1332          * @param $variant
1333          *
1334          * @return string
1335          */
1336         function getRuleConvertedStr( $variant ) {
1337                 $bidtable = $this->mBidtable;
1338                 $unidtable = $this->mUnidtable;
1339
1340                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1341                         return $this->mRules;
1342                 } else {
1343                         // display current variant in bidirectional array
1344                         $disp = $this->getTextInBidtable( $variant );
1345                         // or display current variant in fallbacks
1346                         if ( !$disp ) {
1347                                 $disp = $this->getTextInBidtable(
1348                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1349                         }
1350                         // or display current variant in unidirectional array
1351                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1352                                 $disp = array_values( $unidtable[$variant] );
1353                                 $disp = $disp[0];
1354                         }
1355                         // or display frist text under disable manual convert
1356                         if ( !$disp
1357                                  && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1358                                 if ( count( $bidtable ) > 0 ) {
1359                                         $disp = array_values( $bidtable );
1360                                         $disp = $disp[0];
1361                                 } else {
1362                                         $disp = array_values( $unidtable );
1363                                         $disp = array_values( $disp[0] );
1364                                         $disp = $disp[0];
1365                                 }
1366                         }
1367                         return $disp;
1368                 }
1369         }
1370
1371         /**
1372          * Similar to getRuleConvertedStr(), but this prefers to use original
1373          * page title if $variant === $this->mConverter->mMainLanguageCode
1374          * and may return false in this case (so this title conversion rule
1375          * will be ignored and the original title is shown).
1376          *
1377          * @since 1.22
1378          * @param $variant The variant code to display page title in
1379          * @return String|false The converted title or false if just page name
1380          */
1381         function getRuleConvertedTitle( $variant ) {
1382                 if ( $variant === $this->mConverter->mMainLanguageCode ) {
1383                         // If a string targeting exactly this variant is set,
1384                         // use it. Otherwise, just return false, so the real
1385                         // page name can be shown (and because variant === main,
1386                         // there'll be no further automatic conversion).
1387                         $disp = $this->getTextInBidtable( $variant );
1388                         if ( $disp ) {
1389                                 return $disp;
1390                         }
1391                         if ( array_key_exists( $variant, $this->mUnidtable ) ) {
1392                                 $disp = array_values( $this->mUnidtable[$variant] );
1393                                 $disp = $disp[0];
1394                         }
1395                         // Assigned above or still false.
1396                         return $disp;
1397                 } else {
1398                         return $this->getRuleConvertedStr( $variant );
1399                 }
1400         }
1401
1402         /**
1403          * Generate conversion table for all text.
1404          * @private
1405          */
1406         function generateConvTable() {
1407                 // Special case optimisation
1408                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1409                         $this->mConvTable = array();
1410                         return;
1411                 }
1412
1413                 $bidtable = $this->mBidtable;
1414                 $unidtable = $this->mUnidtable;
1415                 $manLevel = $this->mConverter->mManualLevel;
1416
1417                 $vmarked = array();
1418                 foreach ( $this->mConverter->mVariants as $v ) {
1419                         /* for bidirectional array
1420                                 fill in the missing variants, if any,
1421                                 with fallbacks */
1422                         if ( !isset( $bidtable[$v] ) ) {
1423                                 $variantFallbacks =
1424                                         $this->mConverter->getVariantFallbacks( $v );
1425                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1426                                 if ( $vf ) {
1427                                         $bidtable[$v] = $vf;
1428                                 }
1429                         }
1430
1431                         if ( isset( $bidtable[$v] ) ) {
1432                                 foreach ( $vmarked as $vo ) {
1433                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1434                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1435                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1436                                         // to introduce a custom mapping between
1437                                         // words WordZh and WordTw in the whole text
1438                                         if ( $manLevel[$v] == 'bidirectional' ) {
1439                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1440                                         }
1441                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1442                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1443                                         }
1444                                 }
1445                                 $vmarked[] = $v;
1446                         }
1447                         /* for unidirectional array fill to convert tables */
1448                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1449                                 && isset( $unidtable[$v] ) )
1450                         {
1451                                 if ( isset( $this->mConvTable[$v] ) ) {
1452                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1453                                 } else {
1454                                         $this->mConvTable[$v] = $unidtable[$v];
1455                                 }
1456                         }
1457                 }
1458         }
1459
1460         /**
1461          * Parse rules and flags.
1462          * @param $variant String: variant language code
1463          */
1464         public function parse( $variant = null ) {
1465                 if ( !$variant ) {
1466                         $variant = $this->mConverter->getPreferredVariant();
1467                 }
1468
1469                 $this->parseFlags();
1470                 $flags = $this->mFlags;
1471
1472                 // convert to specified variant
1473                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1474                 if ( $this->mVariantFlags ) {
1475                         // check if current variant in flags
1476                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1477                                 // then convert <text to convert> to current language
1478                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1479                                         $variant );
1480                         } else { // if current variant no in flags,
1481                                    // then we check its fallback variants.
1482                                 $variantFallbacks =
1483                                         $this->mConverter->getVariantFallbacks( $variant );
1484                                 if ( is_array( $variantFallbacks ) ) {
1485                                         foreach ( $variantFallbacks as $variantFallback ) {
1486                                                 // if current variant's fallback exist in flags
1487                                                 if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1488                                                         // then convert <text to convert> to fallback language
1489                                                         $this->mRules =
1490                                                                 $this->mConverter->autoConvert( $this->mRules,
1491                                                                         $variantFallback );
1492                                                         break;
1493                                                 }
1494                                         }
1495                                 }
1496                         }
1497                         $this->mFlags = $flags = array( 'R' => true );
1498                 }
1499
1500                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1501                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1502                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1503                         $this->parseRules();
1504                 }
1505                 $rules = $this->mRules;
1506
1507                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1508                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1509                                 // fill all variants if text in -{A/H/-|text} without rules
1510                                 foreach ( $this->mConverter->mVariants as $v ) {
1511                                         $this->mBidtable[$v] = $rules;
1512                                 }
1513                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1514                                 $this->mFlags = $flags = array( 'R' => true );
1515                         }
1516                 }
1517
1518                 $this->mRuleDisplay = false;
1519                 foreach ( $flags as $flag => $unused ) {
1520                         switch ( $flag ) {
1521                                 case 'R':
1522                                         // if we don't do content convert, still strip the -{}- tags
1523                                         $this->mRuleDisplay = $rules;
1524                                         break;
1525                                 case 'N':
1526                                         // process N flag: output current variant name
1527                                         $ruleVar = trim( $rules );
1528                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1529                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1530                                         } else {
1531                                                 $this->mRuleDisplay = '';
1532                                         }
1533                                         break;
1534                                 case 'D':
1535                                         // process D flag: output rules description
1536                                         $this->mRuleDisplay = $this->getRulesDesc();
1537                                         break;
1538                                 case 'H':
1539                                         // process H,- flag or T only: output nothing
1540                                         $this->mRuleDisplay = '';
1541                                         break;
1542                                 case '-':
1543                                         $this->mRulesAction = 'remove';
1544                                         $this->mRuleDisplay = '';
1545                                         break;
1546                                 case '+':
1547                                         $this->mRulesAction = 'add';
1548                                         $this->mRuleDisplay = '';
1549                                         break;
1550                                 case 'S':
1551                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1552                                         break;
1553                                 case 'T':
1554                                         $this->mRuleTitle = $this->getRuleConvertedTitle( $variant );
1555                                         $this->mRuleDisplay = '';
1556                                         break;
1557                                 default:
1558                                         // ignore unknown flags (but see error case below)
1559                         }
1560                 }
1561                 if ( $this->mRuleDisplay === false ) {
1562                         $this->mRuleDisplay = '<span class="error">'
1563                                 . wfMessage( 'converter-manual-rule-error' )->inContentLanguage()->escaped()
1564                                 . '</span>';
1565                 }
1566
1567                 $this->generateConvTable();
1568         }
1569
1570         /**
1571          * @todo FIXME: code this function :)
1572          */
1573         public function hasRules() {
1574                 // TODO:
1575         }
1576
1577         /**
1578          * Get display text on markup -{...}-
1579          * @return string
1580          */
1581         public function getDisplay() {
1582                 return $this->mRuleDisplay;
1583         }
1584
1585         /**
1586          * Get converted title.
1587          * @return string
1588          */
1589         public function getTitle() {
1590                 return $this->mRuleTitle;
1591         }
1592
1593         /**
1594          * Return how deal with conversion rules.
1595          * @return string
1596          */
1597         public function getRulesAction() {
1598                 return $this->mRulesAction;
1599         }
1600
1601         /**
1602          * Get conversion table. (bidirectional and unidirectional
1603          * conversion table)
1604          * @return array
1605          */
1606         public function getConvTable() {
1607                 return $this->mConvTable;
1608         }
1609
1610         /**
1611          * Get conversion rules string.
1612          * @return string
1613          */
1614         public function getRules() {
1615                 return $this->mRules;
1616         }
1617
1618         /**
1619          * Get conversion flags.
1620          * @return array
1621          */
1622         public function getFlags() {
1623                 return $this->mFlags;
1624         }
1625 }